Beam search

$$\max p(t_1,t_2,\cdots ,t_n)=\max p(t_1)\times p(t_2|t_1)\times \cdots \times p(t_n|t_{n-1}|\cdots |t_2|t_1)$$